NICE Analyzer not collecting 100% call records 1

[Ding07]

We've been using NICE Analyzer for years (since 2001) as an optional tool to analyze calls. Recently we attempted to use it to report on all the calls like other CDR based call detail recording software with no success due to the fact that not 100% of the call records are collected. Is it due to the nature of the NICE Analyzer as a call analysis tool not recording solution, or is this something we may be able to fix through patching, upgrading on both the NICE Analyzer itself and the SQL database server. Currently we use NICE Analyzer 2 with SQL 7.
We notice that the data files are FTPed from CMS to NICE Analyzer every in a 10-minute interval on CMS, is this "batch-mode" data transport (not real-time) somehow causing the loss of call records?
If we really need to report on call record level, should we look for CDR based reporting solutions?

Thank you in advance for your help!

 

[multibus2]

Nice Analyzer (NA) as you found get it's data from the CMS's External Call History data (ECH). The CMS itself only processes ACD measured calls where as CDR will capture all calls if correctly setup. Now the issue is CDR data is quite limted in information compared to the data from the CMS which is very detailed.
So if you want basic call information CDR maybe the choice. If you need detailed information of ACD measured calls NA is the correct choice.
In most Call Centre / general switch setups both CDR and CMS ECH data will use a combination of both data however it should be noted that it's not easy to link CDR data to ECH data directly.


"Been there, done that and got the teeshirt

 

[Ding07]

multibus2,

Thanks very much for your insight! The issue we're having is that NICE Analyzer is missing some call records measured by CMS as ACD calls, for instance, a VDN has about 520 ACD calls a day based on the CMS reports, but we only found 515 NICE Analyzer call records (90-99% but not 100%).
Should every call record be collected, shipped, and loaded into NICE Analyzer SQL database (via CMS ECH process)? If yes, may be something is wrong with our NICE Analyzer and/or related processes on CMS box.

Thanks!

 

[multibus2]

Sure, ECH data is the cradle to grave data from the CMS, as the ACD summary data and ECH data are derived from the same events the number of calls should be the same.
So one of several things are.
Not all the data has been sent, possible, the ECH logs on the CMS should show failed FTP connections.
NA is unable to process some of the data, possible, NA maybe deeming some records as corrupt, the NA logs might show this.
The report criteria in NA does not match the CMS report criteria, also possible.
To be honest without detailed analysis it is difficult to tell.

 

[Ding07]

I checked all the logs, on CMS "/pserv/ech/echlog" and on NICE Analyzer logs, no failed FTP logged. Also, no currupt files are found on NICE Analyser either based on NA log file.
Is it possible that the data some time isn't captured at all?
In other words, the 10-minute FTP session "cvx_ftp" on CMS sends a bacth of data file (e.g. chr0101.772, chr0101.773, etc.) to NICE Analyzer, could it be that some times these file don't even have all the data captured and saved for the calls measured?


Thanks!

 

[Ding07]

I did find one file loading error on NICE Analyzer, it failed to load "synonyms.xml" due to some invalid characters in one of the VDN names. The failed VDN name is not the VDN that is missing call records.
It seems to me that this file includes all the name definitions of agent loginID numbers, VDN names, etc. This file is loaded once every mid-night.

 

[multibus2]

CMS and NA will only record ACD measured Calls. So if these missing calls are not ACD measured calls they will not be recorded. Are you sure the VDN in question is set to be measured to both or external on the CM? If it not set or set to internal there will not be a record on the CMS and thus on the NA. The synonyms.xml is download of all the dictionary items from the CMS.

Hope this helps

"Been there, done that and got the teeshirt

 

[Ding07]

The VDN I've been investigating is measured by BOTH (internally and CMS or externally). The number of call records in NICE Analyzer is always less than the number of ACD calls in CMS VDN reports (interval, daily, etc.) for the same VDN.
I even analyzed the total number of all measured contacts NICE Analyzer has collected for the past 12 days (with total of 150,000 records or calls), and compared this number to the total of all the CMS mearured calls, not 100% match, the accuracy also varies by days.
And I also tried to exclude some items that NICE Analyzer may not be collecting such as calls abandoned in vector ("vecabncalls") but counted in CMS, the overall match got closer but never match exactly.
The most critial number to us is number of ACD calls. But it just does not add up (NICE number is always 1-2% less CMS number).

 

[multibus2]

Interesting. What happens in the associated vector?


"Been there, done that and got the teeshirt

 

[Ding07]

This is the vector:

02 announcement 4321
03 queue-to skill 99 pri m
04 wait-time 12 secs hearing ringback
05 announcement 1233
06 wait-time 24 secs hearing music
07 disconnect after announcement 4142

 

[madonnac]

How many calls are you receiving? 10 minute intervals should be enough for most call centers, but beware of the fact that CMS will roll over in its data collection, possibly overwriting calls that have not yet been collected.

 

[Ding07]

The call volume varies for this particular DNIS, anywhere from 200 up to 1000 calls a day. I have 30-minute interval in CMS. Also, due to the nature of this program (TV/Radio commercial generated), we do experience call spikes.

 

[multibus2]

A nice simple Vector, NA should in theory have all calls routed through it, are they many force disconnected calls?

You could use disp events on the CM to see if it is reporting any interesting events. Other than this it really is difficult to tell.

"Been there, done that and got the teeshirt

 

[Ding07]

There are some foed disconnected calls on this DNIS if no agents are staffed.
But either forced disconnect calls or ACD calls simply do not add up for some intervals. For example, NICE Analyzer has 4 call records for an agent between 10:00AM and 10:30AM but CMS "hagent" table has 5 ACD calls for the that agent.

 

[multibus2]

Are the calls missing or does the previous or next interval have more call? Without detailed diagnosis it's impossible to tell.


"Been there, done that and got the teeshirt

 

[Ding07]

Yes, some call records are missing in NICE Analyzer. I did check the previous or next interval.

 

[multibus2]

Okay. To be honest without a detailed diagnosis it will impossibile to tell.

"Been there, done that and got the teeshirt